cellects 0.1.0.dev1__py3-none-any.whl

cellects/core/cellects_threads.py

@@ -0,0 +1,1464 @@
+#!/usr/bin/env python3
+"""
+Cellects graphical user interface interacts with computational scripts through threads.
+Especially, each thread calls one or several methods of the class named "program_organizer",
+which regroup all available computation of the software.
+These threads are started from a children of WindowType, run methods from program_organizer and send messages and
+results to the corresponding children of WindowType, allowing, for instance, to display a result in the interface.
+"""
+
+import logging
+import weakref
+from multiprocessing import Queue, Process, Manager
+import os
+import time
+from glob import glob
+from timeit import default_timer
+from copy import deepcopy
+import cv2
+from numba.typed import Dict as TDict
+import numpy as np
+import pandas as pd
+from PySide6 import QtCore
+from cellects.image_analysis.morphological_operations import cross_33, Ellipse
+from cellects.image_analysis.image_segmentation import generate_color_space_combination
+from cellects.utils.load_display_save import read_and_rotate
+from cellects.utils.utilitarian import PercentAndTimeTracker, reduce_path_len
+from cellects.core.one_video_per_blob import OneVideoPerBlob
+from cellects.utils.load_display_save import write_video
+from cellects.core.motion_analysis import MotionAnalysis
+
+
+class LoadDataToRunCellectsQuicklyThread(QtCore.QThread):
+    message_from_thread = QtCore.Signal(str)
+
+    def __init__(self, parent=None):
+        super(LoadDataToRunCellectsQuicklyThread, self).__init__(parent)
+        self.setParent(parent)
+
+    def run(self):
+        self.parent().po.look_for_data()
+        self.parent().po.load_data_to_run_cellects_quickly()
+        if self.parent().po.first_exp_ready_to_run:
+            self.message_from_thread.emit("Data found, Video tracking window and Run all directly are available")
+        else:
+            self.message_from_thread.emit("")
+
+
+class LookForDataThreadInFirstW(QtCore.QThread):
+    def __init__(self, parent=None):
+        super(LookForDataThreadInFirstW, self).__init__(parent)
+        self.setParent(parent)
+
+    def run(self):
+        self.parent().po.look_for_data()
+
+
+class LoadFirstFolderIfSeveralThread(QtCore.QThread):
+    message_when_thread_finished = QtCore.Signal(bool)
+    def __init__(self, parent=None):
+        super(LoadFirstFolderIfSeveralThread, self).__init__(parent)
+        self.setParent(parent)
+
+    def run(self):
+        self.parent().po.load_data_to_run_cellects_quickly()
+        if not self.parent().po.first_exp_ready_to_run:
+            self.parent().po.get_first_image()
+        self.message_when_thread_finished.emit(self.parent().po.first_exp_ready_to_run)
+
+
+class GetFirstImThread(QtCore.QThread):
+    message_when_thread_finished = QtCore.Signal(bool)
+    def __init__(self, parent=None):
+        """
+        This class read the first image of the (first of the) selected analysis.
+        According to the first_detection_frame value,it can be another image
+        If this is the first time a first image is read, it also gather the following variables:
+            - img_number
+            - dims (video dimensions: time, y, x)
+            - raw_images (whether images are in a raw format)
+        If the selected analysis contains videos instead of images, it opens the first video
+        and read the first_detection_frame th image.
+        :param parent: An object containing all necessary variables.
+        """
+        super(GetFirstImThread, self).__init__(parent)
+        self.setParent(parent)
+
+    def run(self):
+        self.parent().po.get_first_image()
+        self.message_when_thread_finished.emit(True)
+
+
+class GetLastImThread(QtCore.QThread):
+    def __init__(self, parent=None):
+        super(GetLastImThread, self).__init__(parent)
+        self.setParent(parent)
+
+    def run(self):
+        self.parent().po.get_last_image()
+
+
+class UpdateImageThread(QtCore.QThread):
+    message_when_thread_finished = QtCore.Signal(bool)
+
+    def __init__(self, parent=None):
+        super(UpdateImageThread, self).__init__(parent)
+        self.setParent(parent)
+
+    def run(self):
+        # I/ If this thread runs from user input, get the right coordinates
+        # and convert them to fit the displayed image size
+        user_input = len(self.parent().imageanalysiswindow.saved_coord) > 0 or len(self.parent().imageanalysiswindow.temporary_mask_coord) > 0
+        if user_input:
+            if len(self.parent().imageanalysiswindow.temporary_mask_coord) > 0:
+                idx = self.parent().imageanalysiswindow.temporary_mask_coord
+            else:
+                idx = self.parent().imageanalysiswindow.saved_coord
+            if len(idx) < 2:
+                user_input = False
+            else:
+                # Convert coordinates:
+                self.parent().imageanalysiswindow.display_image.update_image_scaling_factors()
+                sf = self.parent().imageanalysiswindow.display_image.scaling_factors
+                idx = np.array(((np.round(idx[0][0] * sf[0]), np.round(idx[0][1] * sf[1])), (np.round(idx[1][0] * sf[0]), np.round(idx[1][1] * sf[1]))), dtype=np.int64)
+                min_y = np.min(idx[:, 0])
+                max_y = np.max(idx[:, 0])
+                min_x = np.min(idx[:, 1])
+                max_x = np.max(idx[:, 1])
+                if max_y > self.parent().imageanalysiswindow.drawn_image.shape[0]:
+                    max_y = self.parent().imageanalysiswindow.drawn_image.shape[0] - 1
+                if max_x > self.parent().imageanalysiswindow.drawn_image.shape[1]:
+                    max_x = self.parent().imageanalysiswindow.drawn_image.shape[1] - 1
+                if min_y < 0:
+                    min_y = 0
+                if min_x < 0:
+                    min_x = 0
+
+        if len(self.parent().imageanalysiswindow.temporary_mask_coord) == 0:
+            # not_load
+            # II/ If this thread aims at saving the last user input and displaying all user inputs:
+            # Update the drawn_image according to every saved masks
+            # 1) The segmentation mask
+            # 2) The back_mask and bio_mask
+            # 3) The automatically detected video contours
+            # (re-)Initialize drawn image
+            self.parent().imageanalysiswindow.drawn_image = deepcopy(self.parent().po.current_image)
+            if self.parent().imageanalysiswindow.drawn_image.size < 1000000:
+                contour_width = 3
+            else:
+                contour_width = 6
+            # 1) The segmentation mask
+            logging.info('Add the segmentation mask to the image')
+            if self.parent().imageanalysiswindow.is_first_image_flag:
+                im_combinations = self.parent().po.first_image.im_combinations
+                im_mean = self.parent().po.first_image.image.mean()
+            else:
+                im_combinations = self.parent().po.last_image.im_combinations
+                im_mean = self.parent().po.last_image.bgr.mean()
+            # If there are image combinations, get the current corresponding binary image
+            if im_combinations is not None and len(im_combinations) != 0:
+                binary_idx = im_combinations[self.parent().po.current_combination_id]["binary_image"]
+                # If it concerns the last image, only keep the contour coordinates
+
+                cv2.eroded_binary = cv2.erode(binary_idx, cross_33)
+                binary_idx = binary_idx - cv2.eroded_binary
+                binary_idx = cv2.dilate(binary_idx, kernel=cross_33, iterations=contour_width)
+                binary_idx = np.nonzero(binary_idx)
+                # Color these coordinates in magenta on bright images, and in pink on dark images
+                if im_mean > 126:
+                    # logging.info('Color the segmentation mask in magenta')
+                    self.parent().imageanalysiswindow.drawn_image[binary_idx[0], binary_idx[1], :] = np.array((20, 0, 150), dtype=np.uint8)
+                else:
+                    # logging.info('Color the segmentation mask in pink')
+                    self.parent().imageanalysiswindow.drawn_image[binary_idx[0], binary_idx[1], :] = np.array((94, 0, 213), dtype=np.uint8)
+            if user_input:# save
+                mask = np.zeros(self.parent().imageanalysiswindow.drawn_image.shape[:2], dtype=np.uint8)
+                if self.parent().imageanalysiswindow.back1_bio2 == 0:
+                    logging.info("Save the user drawn mask of the current arena")
+                    if self.parent().po.vars['arena_shape'] == 'circle':
+                        ellipse = Ellipse((max_y - min_y, max_x - min_x)).create().astype(np.uint8)
+                        mask[min_y:max_y, min_x:max_x, ...] = ellipse
+                    else:
+                        mask[min_y:max_y, min_x:max_x] = 1
+                else:
+                    logging.info("Save the user drawn mask of Cell or Back")
+
+                    if self.parent().imageanalysiswindow.back1_bio2 == 2:
+                        if self.parent().po.all['starting_blob_shape'] == 'circle':
+                            ellipse = Ellipse((max_y - min_y, max_x - min_x)).create().astype(np.uint8)
+                            mask[min_y:max_y, min_x:max_x, ...] = ellipse
+                        else:
+                            mask[min_y:max_y, min_x:max_x] = 1
+                    else:
+                        mask[min_y:max_y, min_x:max_x] = 1
+                mask = np.nonzero(mask)
+
+                if self.parent().imageanalysiswindow.back1_bio2 == 1:
+                    self.parent().imageanalysiswindow.back_masks_number += 1
+                    self.parent().imageanalysiswindow.back_mask[mask[0], mask[1]] = self.parent().imageanalysiswindow.available_back_names[0]
+                elif self.parent().imageanalysiswindow.back1_bio2 == 2:
+                    self.parent().imageanalysiswindow.bio_masks_number += 1
+                    self.parent().imageanalysiswindow.bio_mask[mask[0], mask[1]] = self.parent().imageanalysiswindow.available_bio_names[0]
+                elif self.parent().imageanalysiswindow.manual_delineation_flag:
+                    self.parent().imageanalysiswindow.arena_masks_number += 1
+                    self.parent().imageanalysiswindow.arena_mask[mask[0], mask[1]] = self.parent().imageanalysiswindow.available_arena_names[0]
+                # 2)a) Apply all these masks to the drawn image:
+
+            back_coord = np.nonzero(self.parent().imageanalysiswindow.back_mask)
+
+            bio_coord = np.nonzero(self.parent().imageanalysiswindow.bio_mask)
+
+            if self.parent().imageanalysiswindow.arena_mask is not None:
+                arena_coord = np.nonzero(self.parent().imageanalysiswindow.arena_mask)
+                self.parent().imageanalysiswindow.drawn_image[arena_coord[0], arena_coord[1], :] = np.repeat(self.parent().po.vars['contour_color'], 3).astype(np.uint8)
+
+            self.parent().imageanalysiswindow.drawn_image[back_coord[0], back_coord[1], :] = np.array((224, 160, 81), dtype=np.uint8)
+
+            self.parent().imageanalysiswindow.drawn_image[bio_coord[0], bio_coord[1], :] = np.array((17, 160, 212), dtype=np.uint8)
+
+            image = self.parent().imageanalysiswindow.drawn_image
+            # 3) The automatically detected video contours
+            if self.parent().imageanalysiswindow.delineation_done:  # add a mask of the video contour
+                # logging.info("Draw the delineation mask of each arena")
+                for contour_i in range(len(self.parent().po.top)):
+                    mask = np.zeros(self.parent().imageanalysiswindow.drawn_image.shape[:2], dtype=np.uint8)
+                    min_cy = self.parent().po.top[contour_i]
+                    max_cy = self.parent().po.bot[contour_i]
+                    min_cx = self.parent().po.left[contour_i]
+                    max_cx = self.parent().po.right[contour_i]
+                    text = f"{contour_i + 1}"
+                    position = (self.parent().po.left[contour_i] + 25, self.parent().po.top[contour_i] + (self.parent().po.bot[contour_i] - self.parent().po.top[contour_i]) // 2)
+                    image = cv2.putText(image,  # numpy array on which text is written
+                                    text,  # text
+                                    position,  # position at which writing has to start
+                                    cv2.FONT_HERSHEY_SIMPLEX,  # font family
+                                    1,  # font size
+                                    (138, 95, 18, 255),
+                                    # (209, 80, 0, 255),  # font color
+                                    2)  # font stroke
+                    if (max_cy - min_cy) < 0 or (max_cx - min_cx) < 0:
+                        self.parent().imageanalysiswindow.message.setText("Error: the shape number or the detection is wrong")
+                    if self.parent().po.vars['arena_shape'] == 'circle':
+                        ellipse = Ellipse((max_cy - min_cy, max_cx - min_cx)).create().astype(np.uint8)
+                        ellipse = cv2.morphologyEx(ellipse, cv2.MORPH_GRADIENT, cross_33)
+                        mask[min_cy:max_cy, min_cx:max_cx, ...] = ellipse
+                    else:
+                        mask[(min_cy, max_cy), min_cx:max_cx] = 1
+                        mask[min_cy:max_cy, (min_cx, max_cx)] = 1
+                    mask = cv2.dilate(mask, kernel=cross_33, iterations=contour_width)
+
+                    mask = np.nonzero(mask)
+                    image[mask[0], mask[1], :] = np.array((138, 95, 18), dtype=np.uint8)# self.parent().po.vars['contour_color']
+
+        else: #load
+            if user_input:
+                # III/ If this thread runs from user input: update the drawn_image according to the current user input
+                # Just add the mask to drawn_image as quick as possible
+                # Add user defined masks
+                # Take the drawn image and add the temporary mask to it
+                image = deepcopy(self.parent().imageanalysiswindow.drawn_image)
+                if self.parent().imageanalysiswindow.back1_bio2 == 0:
+                    # logging.info("Dynamic drawing of the arena outline")
+                    if self.parent().po.vars['arena_shape'] == 'circle':
+                        ellipse = Ellipse((max_y - min_y, max_x - min_x)).create()
+                        ellipse = np.stack((ellipse, ellipse, ellipse), axis=2).astype(np.uint8)
+                        image[min_y:max_y, min_x:max_x, ...] *= (1 - ellipse)
+                        image[min_y:max_y, min_x:max_x, ...] += ellipse
+                    else:
+                        mask = np.zeros(self.parent().imageanalysiswindow.drawn_image.shape[:2], dtype=np.uint8)
+                        mask[min_y:max_y, min_x:max_x] = 1
+                        mask = np.nonzero(mask)
+                        image[mask[0], mask[1], :] = np.array((0, 0, 0), dtype=np.uint8)
+                else:
+                    # logging.info("Dynamic drawing of Cell or Back")
+                    if self.parent().imageanalysiswindow.back1_bio2 == 2:
+                        if self.parent().po.all['starting_blob_shape'] == 'circle':
+                            ellipse = Ellipse((max_y - min_y, max_x - min_x)).create()
+                            ellipse = np.stack((ellipse, ellipse, ellipse), axis=2).astype(np.uint8)
+                            image[min_y:max_y, min_x:max_x, ...] *= (1 - ellipse)
+                            ellipse[:, :, :] *= np.array((17, 160, 212), dtype=np.uint8)
+                            image[min_y:max_y, min_x:max_x, ...] += ellipse
+                        else:
+                            mask = np.zeros(self.parent().imageanalysiswindow.drawn_image.shape[:2], dtype=np.uint8)
+                            mask[min_y:max_y, min_x:max_x] = 1
+                            mask = np.nonzero(mask)
+                            image[mask[0], mask[1], :] = np.array((17, 160, 212), dtype=np.uint8)
+                    else:
+                        mask = np.zeros(self.parent().imageanalysiswindow.drawn_image.shape[:2], dtype=np.uint8)
+                        mask[min_y:max_y, min_x:max_x] = 1
+                        mask = np.nonzero(mask)
+                        image[mask[0], mask[1], :] = np.array((224, 160, 81), dtype=np.uint8)
+
+        self.parent().imageanalysiswindow.display_image.update_image(image)
+        self.message_when_thread_finished.emit(True)
+
+
+class FirstImageAnalysisThread(QtCore.QThread):
+    message_from_thread = QtCore.Signal(str)
+    message_when_thread_finished = QtCore.Signal(bool)
+
+    def __init__(self, parent=None):
+        super(FirstImageAnalysisThread, self).__init__(parent)
+        self.setParent(parent)
+
+    def run(self):
+        tic = default_timer()
+        biomask = None
+        backmask = None
+        if self.parent().imageanalysiswindow.bio_masks_number != 0:
+            shape_nb, ordered_image = cv2.connectedComponents((self.parent().imageanalysiswindow.bio_mask > 0).astype(np.uint8))
+            shape_nb -= 1
+            biomask = np.nonzero(self.parent().imageanalysiswindow.bio_mask)
+        else:
+            shape_nb = 0
+        if self.parent().imageanalysiswindow.back_masks_number != 0:
+            backmask = np.nonzero(self.parent().imageanalysiswindow.back_mask)
+        if self.parent().po.visualize or len(self.parent().po.first_im.shape) == 2 or shape_nb == self.parent().po.sample_number:
+            self.message_from_thread.emit("Image segmentation, wait 30 seconds at most")
+            if not self.parent().imageanalysiswindow.asking_first_im_parameters_flag and self.parent().po.all['scale_with_image_or_cells'] == 0 and self.parent().po.all["set_spot_size"]:
+                self.parent().po.get_average_pixel_size()
+                spot_size = self.parent().po.starting_blob_hsize_in_pixels
+            else:
+                spot_size = None
+            self.parent().po.all["bio_mask"] = biomask
+            self.parent().po.all["back_mask"] = backmask
+            self.parent().po.fast_image_segmentation(is_first_image=True, biomask=biomask, backmask=backmask, spot_size=spot_size)
+            if shape_nb == self.parent().po.sample_number and self.parent().po.first_image.im_combinations[self.parent().po.current_combination_id]['shape_number'] != self.parent().po.sample_number:
+                self.parent().po.first_image.im_combinations[self.parent().po.current_combination_id]['shape_number'] = shape_nb
+                self.parent().po.first_image.shape_number = shape_nb
+                self.parent().po.first_image.validated_shapes = (self.parent().imageanalysiswindow.bio_mask > 0).astype(np.uint8)
+                self.parent().po.first_image.im_combinations[self.parent().po.current_combination_id]['binary_image'] = self.parent().po.first_image.validated_shapes
+        else:
+            self.message_from_thread.emit("Generating analysis options, wait...")
+            if self.parent().po.vars["color_number"] > 2:
+                kmeans_clust_nb = self.parent().po.vars["color_number"]
+                if self.parent().po.carefully:
+                    self.message_from_thread.emit("Generating analysis options, wait less than 30 minutes")
+                else:
+                    self.message_from_thread.emit("Generating analysis options, a few minutes")
+            else:
+                kmeans_clust_nb = None
+                if self.parent().po.carefully:
+                    self.message_from_thread.emit("Generating analysis options, wait a few minutes")
+                else:
+                    self.message_from_thread.emit("Generating analysis options, around 1 minute")
+            if self.parent().imageanalysiswindow.asking_first_im_parameters_flag:
+                self.parent().po.first_image.find_first_im_csc(sample_number=self.parent().po.sample_number,
+                                                               several_blob_per_arena=None,
+                                                               spot_shape=None, spot_size=None,
+                                                               kmeans_clust_nb=kmeans_clust_nb,
+                                                               biomask=self.parent().po.all["bio_mask"],
+                                                               backmask=self.parent().po.all["back_mask"],
+                                                               color_space_dictionaries=None,
+                                                               carefully=self.parent().po.carefully)
+            else:
+                if self.parent().po.all['scale_with_image_or_cells'] == 0:
+                    self.parent().po.get_average_pixel_size()
+                else:
+                    self.parent().po.starting_blob_hsize_in_pixels = None
+                self.parent().po.first_image.find_first_im_csc(sample_number=self.parent().po.sample_number,
+                                                                                   several_blob_per_arena=self.parent().po.vars['several_blob_per_arena'],
+                                                                                   spot_shape=self.parent().po.all['starting_blob_shape'],
+                                                               spot_size=self.parent().po.starting_blob_hsize_in_pixels,
+                                                                                   kmeans_clust_nb=kmeans_clust_nb,
+                                                                                   biomask=self.parent().po.all["bio_mask"],
+                                                                                   backmask=self.parent().po.all["back_mask"],
+                                                                                   color_space_dictionaries=None,
+                                                                                   carefully=self.parent().po.carefully)
+
+        logging.info(f" image analysis lasted {default_timer() - tic} secondes")
+        logging.info(f" image analysis lasted {np.round((default_timer() - tic) / 60)} minutes")
+        self.message_when_thread_finished.emit(True)
+
+
+class LastImageAnalysisThread(QtCore.QThread):
+    message_from_thread = QtCore.Signal(str)
+    message_when_thread_finished = QtCore.Signal(bool)
+
+    def __init__(self, parent=None):
+        super(LastImageAnalysisThread, self).__init__(parent)
+        self.setParent(parent)
+
+    def run(self):
+        self.parent().po.cropping(False)
+        self.parent().po.get_background_to_subtract()
+        biomask = None
+        backmask = None
+        if self.parent().imageanalysiswindow.bio_masks_number != 0:
+            biomask = np.nonzero(self.parent().imageanalysiswindow.bio_mask)
+        if self.parent().imageanalysiswindow.back_masks_number != 0:
+            backmask = np.nonzero(self.parent().imageanalysiswindow.back_mask)
+        if self.parent().po.visualize or len(self.parent().po.first_im.shape) == 2:
+            self.message_from_thread.emit("Image segmentation, wait...")
+            self.parent().po.fast_image_segmentation(is_first_image=False, biomask=biomask, backmask=backmask)
+        else:
+            self.message_from_thread.emit("Generating analysis options, wait...")
+            if self.parent().po.vars['several_blob_per_arena']:
+                concomp_nb = [self.parent().po.sample_number, self.parent().po.first_image.size // 50]
+                max_shape_size = .75 * self.parent().po.first_image.size
+                total_surfarea = .99 * self.parent().po.first_image.size
+            else:
+                concomp_nb = [self.parent().po.sample_number, self.parent().po.sample_number * 200]
+                if self.parent().po.all['are_zigzag'] == "columns":
+                    inter_dist = np.mean(np.diff(np.nonzero(self.parent().po.videos.first_image.y_boundaries)))
+                elif self.parent().po.all['are_zigzag'] == "rows":
+                    inter_dist = np.mean(np.diff(np.nonzero(self.parent().po.videos.first_image.x_boundaries)))
+                else:
+                    dist1 = np.mean(np.diff(np.nonzero(self.parent().po.videos.first_image.y_boundaries)))
+                    dist2 = np.mean(np.diff(np.nonzero(self.parent().po.videos.first_image.x_boundaries)))
+                    inter_dist = np.max(dist1, dist2)
+                if self.parent().po.all['starting_blob_shape'] == "circle":
+                    max_shape_size = np.pi * np.square(inter_dist)
+                else:
+                    max_shape_size = np.square(2 * inter_dist)
+                total_surfarea = max_shape_size * self.parent().po.sample_number
+            out_of_arenas = None
+            if self.parent().po.all['are_gravity_centers_moving'] != 1:
+                out_of_arenas = np.ones_like(self.parent().po.videos.first_image.validated_shapes)
+                for blob_i in np.arange(len(self.parent().po.vars['analyzed_individuals'])):
+                    out_of_arenas[self.parent().po.top[blob_i]: (self.parent().po.bot[blob_i] + 1),
+                    self.parent().po.left[blob_i]: (self.parent().po.right[blob_i] + 1)] = 0
+            ref_image = self.parent().po.first_image.validated_shapes
+            self.parent().po.first_image.generate_subtract_background(self.parent().po.vars['convert_for_motion'])
+            kmeans_clust_nb = None
+            self.parent().po.last_image.find_last_im_csc(concomp_nb, total_surfarea, max_shape_size, out_of_arenas,
+                                                         ref_image, self.parent().po.first_image.subtract_background,
+                                                         kmeans_clust_nb, biomask, backmask, color_space_dictionaries=None,
+                                                         carefully=self.parent().po.carefully)
+        self.message_when_thread_finished.emit(True)
+
+
+class CropScaleSubtractDelineateThread(QtCore.QThread):
+    message_from_thread = QtCore.Signal(str)
+    message_when_thread_finished = QtCore.Signal(str)
+
+    def __init__(self, parent=None):
+        super(CropScaleSubtractDelineateThread, self).__init__(parent)
+        self.setParent(parent)
+
+    def run(self):
+        logging.info("Start cropping if required")
+        self.parent().po.cropping(is_first_image=True)
+        self.parent().po.cropping(is_first_image=False)
+        self.parent().po.get_average_pixel_size()
+        if os.path.isfile('Data to run Cellects quickly.pkl'):
+            os.remove('Data to run Cellects quickly.pkl')
+        logging.info("Save data to run Cellects quickly")
+        self.parent().po.data_to_save['first_image'] = True
+        self.parent().po.save_data_to_run_cellects_quickly()
+        self.parent().po.data_to_save['first_image'] = False
+        if not self.parent().po.vars['several_blob_per_arena']:
+            logging.info("Check whether the detected shape number is ok")
+            nb, shapes, stats, centroids = cv2.connectedComponentsWithStats(self.parent().po.first_image.validated_shapes)
+            y_lim = self.parent().po.first_image.y_boundaries
+            if ((nb - 1) != self.parent().po.sample_number or np.any(stats[:, 4] == 1)):
+                self.message_from_thread.emit("Image analysis failed to detect the right cell(s) number: restart the analysis.")
+            elif len(np.nonzero(y_lim == - 1)) != len(np.nonzero(y_lim == 1)):
+                self.message_from_thread.emit("Automatic arena delineation cannot work if one cell touches the image border.")
+                self.parent().po.first_image.y_boundaries = None
+            else:
+                logging.info("Start automatic video delineation")
+                analysis_status = self.parent().po.delineate_each_arena()
+                self.message_when_thread_finished.emit(analysis_status["message"])
+        else:
+            logging.info("Start automatic video delineation")
+            analysis_status = self.parent().po.delineate_each_arena()
+            self.message_when_thread_finished.emit(analysis_status["message"])
+
+
+class SaveManualDelineationThread(QtCore.QThread):
+
+    def __init__(self, parent=None):
+        super(SaveManualDelineationThread, self).__init__(parent)
+        self.setParent(parent)
+
+    def run(self):
+        self.parent().po.left = np.arange(self.parent().po.sample_number)
+        self.parent().po.right = np.arange(self.parent().po.sample_number)
+        self.parent().po.top = np.arange(self.parent().po.sample_number)
+        self.parent().po.bot = np.arange(self.parent().po.sample_number)
+        for arena in np.arange(1, self.parent().po.sample_number + 1):
+            y, x = np.nonzero(self.parent().imageanalysiswindow.arena_mask == arena)
+            self.parent().po.left[arena - 1] = np.min(x)
+            self.parent().po.right[arena - 1] = np.max(x)
+            self.parent().po.top[arena - 1] = np.min(y)
+            self.parent().po.bot[arena - 1] = np.max(y)
+
+        logging.info("Save data to run Cellects quickly")
+        self.parent().po.data_to_save['coordinates'] = True
+        self.parent().po.save_data_to_run_cellects_quickly()
+        self.parent().po.data_to_save['coordinates'] = False
+
+        logging.info("Save manual video delineation")
+        self.parent().po.vars['analyzed_individuals'] = np.arange(self.parent().po.sample_number) + 1
+        self.parent().po.videos = OneVideoPerBlob(self.parent().po.first_image, self.parent().po.starting_blob_hsize_in_pixels, self.parent().po.all['raw_images'])
+        self.parent().po.videos.left = self.parent().po.left
+        self.parent().po.videos.right = self.parent().po.right
+        self.parent().po.videos.top = self.parent().po.top
+        self.parent().po.videos.bot = self.parent().po.bot
+
+
+class GetExifDataThread(QtCore.QThread):
+
+    def __init__(self, parent=None):
+        super(GetExifDataThread, self).__init__(parent)
+        self.setParent(parent)
+
+    def run(self):
+        self.parent().po.extract_exif()
+
+
+class FinalizeImageAnalysisThread(QtCore.QThread):
+
+    def __init__(self, parent=None):
+        super(FinalizeImageAnalysisThread, self).__init__(parent)
+        self.setParent(parent)
+
+    def run(self):
+        self.parent().po.get_background_to_subtract()
+
+        self.parent().po.get_origins_and_backgrounds_lists()
+
+        if self.parent().po.last_image is None:
+            self.parent().po.get_last_image()
+            self.parent().po.fast_image_segmentation(False)
+        self.parent().po.find_if_lighter_background()
+        logging.info("The current (or the first) folder is ready to run")
+        self.parent().po.first_exp_ready_to_run = True
+        self.parent().po.data_to_save['coordinates'] = True
+        self.parent().po.data_to_save['exif'] = True
+        self.parent().po.save_data_to_run_cellects_quickly()
+        self.parent().po.data_to_save['coordinates'] = False
+        self.parent().po.data_to_save['exif'] = False
+
+
+class SaveAllVarsThread(QtCore.QThread):
+
+    def __init__(self, parent=None):
+        super(SaveAllVarsThread, self).__init__(parent)
+        self.setParent(parent)
+
+    def run(self):
+        self.parent().po.save_variable_dict()
+
+        #self.parent().po.all['global_pathway']
+        #os.getcwd()
+
+        self.set_current_folder()
+        self.parent().po.save_data_to_run_cellects_quickly(new_one_if_does_not_exist=False)
+        #if os.access(f"", os.R_OK):
+        #    self.parent().po.save_data_to_run_cellects_quickly()
+        #else:
+        #    logging.error(f"No permission access to write in {os.getcwd()}")
+
+    def set_current_folder(self):
+        if self.parent().po.all['folder_number'] > 1: # len(self.parent().po.all['folder_list']) > 1:  # len(self.parent().po.all['folder_list']) > 0:
+            logging.info(f"Use {self.parent().po.all['folder_list'][0]} folder")
+            self.parent().po.update_folder_id(self.parent().po.all['sample_number_per_folder'][0],
+                                              self.parent().po.all['folder_list'][0])
+        else:
+            curr_path = reduce_path_len(self.parent().po.all['global_pathway'], 6, 10)
+            logging.info(f"Use {curr_path} folder")
+            self.parent().po.update_folder_id(self.parent().po.all['first_folder_sample_number'])
+
+
+class OneArenaThread(QtCore.QThread):
+    message_from_thread_starting = QtCore.Signal(str)
+    image_from_thread = QtCore.Signal(dict)
+    when_loading_finished = QtCore.Signal(bool)
+    when_detection_finished = QtCore.Signal(str)
+
+    def __init__(self, parent=None):
+        super(OneArenaThread, self).__init__(parent)
+        self.setParent(parent)
+        self._isRunning = False
+
+    def run(self):
+        continue_analysis = True
+        self._isRunning = True
+        self.message_from_thread_starting.emit("Video loading, wait...")
+
+        self.set_current_folder() #DOIT être fait à chaque écriture
+        print(self.parent().po.vars['convert_for_motion'])
+        if not self.parent().po.first_exp_ready_to_run:
+            self.parent().po.load_data_to_run_cellects_quickly()
+            if not self.parent().po.first_exp_ready_to_run:
+                #Need a look for data when Data to run Cellects quickly.pkl and 1 folder selected amon several
+                continue_analysis = self.pre_processing()
+        if continue_analysis:
+            print(self.parent().po.vars['convert_for_motion'])
+            memory_diff = self.parent().po.update_available_core_nb()
+            if self.parent().po.cores == 0:
+                self.message_from_thread_starting.emit(f"Analyzing one arena requires {memory_diff}GB of additional RAM to run")
+            else:
+                if self.parent().po.motion is None or self.parent().po.load_quick_full == 0:
+                    self.load_one_arena()
+                if self.parent().po.load_quick_full > 0:
+                    if self.parent().po.motion.start is not None:
+                        logging.info("One arena detection has started")
+                        self.detection()
+                        if self.parent().po.load_quick_full > 1:
+                            logging.info("One arena post-processing has started")
+                            self.post_processing()
+                        else:
+                            self.when_detection_finished.emit("Detection done, read to see the result")
+                    else:
+                        self.message_from_thread_starting.emit(f"The current parameters failed to detect the cell(s) motion")
+
+    def stop(self):
+        self._isRunning = False
+
+    def set_current_folder(self):
+        # if isinstance(self.parent().po.all['sample_number_per_folder'], int):
+        #     self.parent().po.all['folder_number'] = 1
+        # self.parent().po.look_for_data()
+        if self.parent().po.all['folder_number'] > 1: # len(self.parent().po.all['folder_list']) > 1:  # len(self.parent().po.all['folder_list']) > 0:
+            logging.info(f"Use {self.parent().po.all['folder_list'][0]} folder")
+            self.parent().po.update_folder_id(self.parent().po.all['sample_number_per_folder'][0],
+                                              self.parent().po.all['folder_list'][0])
+        else:
+            curr_path = reduce_path_len(self.parent().po.all['global_pathway'], 6, 10)
+            logging.info(f"Use {curr_path} folder")
+            self.parent().po.update_folder_id(self.parent().po.all['first_folder_sample_number'])
+        # logging.info("Look for images/videos data")
+        # self.parent().po.look_for_data()
+        # if len(self.parent().po.all['folder_list']) > 1:  # len(self.parent().po.all['folder_list']) > 0:
+        #     logging.info("Update sub-folder")
+        #     self.parent().po.update_folder_id(self.parent().po.all['sample_number_per_folder'][0],
+        #                                       self.parent().po.all['folder_list'][0])
+        # else:
+        #     self.parent().po.update_folder_id(self.parent().po.all['first_folder_sample_number'])
+
+    def pre_processing(self):
+        logging.info("Pre-processing has started")
+        analysis_status = {"continue": True, "message": ""}
+        # Thinds to save and load here:
+        # situations : 1° on a tout fait et osed de succeed to load
+        # if not self.parent().po.first_exp_ready_to_run réglé
+        # 2° on a rien fait et succeed
+        # 3° on a rien fait et non succeed
+        # 4° Ce dosser n'est pas le premier'
+        # if self.parent().po.succeed_to_data_to_run_cellects_quickly:
+        #     self.message_from_thread_starting.emit(f"Do image analysis first, by clicking Next on the first window")
+        # if not self.parent().po.all['overwrite_cellects_data'] and os.path.isfile(f'Data to run Cellects quickly.pkl'):
+        #     success = self.parent().po.load_data_to_run_cellects_quickly()
+        #     if not success:
+        #         self.message_from_thread_starting.emit(f"Do image analysis first, by clicking Next on the first window")
+        # else:
+
+        self.parent().po.get_first_image()
+        self.parent().po.fast_image_segmentation(is_first_image=True)
+        if len(self.parent().po.vars['analyzed_individuals']) != self.parent().po.first_image.shape_number:
+            self.message_from_thread_starting.emit(f"Wrong specimen number: (re)do the complete analysis.")
+            analysis_status["continue"] = False
+        else:
+            self.parent().po.cropping(is_first_image=True)
+            self.parent().po.get_average_pixel_size()
+            analysis_status = self.parent().po.delineate_each_arena()
+            if not analysis_status["continue"]:
+                self.message_from_thread_starting.emit(analysis_status["message"])
+                logging.error(analysis_status['message'])
+            else:
+                self.parent().po.data_to_save['exif'] = True
+                self.parent().po.save_data_to_run_cellects_quickly()
+                self.parent().po.data_to_save['exif'] = False
+                # self.parent().po.extract_exif()
+                self.parent().po.get_background_to_subtract()
+                if len(self.parent().po.vars['analyzed_individuals']) != len(self.parent().po.top):
+                    self.message_from_thread_starting.emit(f"Wrong specimen number: (re)do the complete analysis.")
+                    analysis_status["continue"] = False
+                else:
+                    self.parent().po.get_origins_and_backgrounds_lists()
+                    self.parent().po.get_last_image()
+                    self.parent().po.fast_image_segmentation(False)
+                    # self.parent().po.type_csc_dict()
+                    self.parent().po.find_if_lighter_backgnp.round()
+                    logging.info("The current (or the first) folder is ready to run")
+                    self.parent().po.first_exp_ready_to_run = True
+        return analysis_status["continue"]
+
+    def load_one_arena(self):
+        arena = self.parent().po.all['arena']
+        i = np.nonzero(self.parent().po.vars['analyzed_individuals'] == arena)[0][0]
+        save_loaded_video: bool = False
+        if not os.path.isfile(f'ind_{arena}.npy') or self.parent().po.all['overwrite_unaltered_videos']:
+            logging.info(f"Starting to load arena n°{arena} from images")
+            add_to_c = 1
+            self.parent().po.one_arenate_done = True
+            i = np.nonzero(self.parent().po.vars['analyzed_individuals'] == arena)[0][0]
+            if self.parent().po.vars['lose_accuracy_to_save_memory']:
+                self.parent().po.converted_video = np.zeros(
+                    (len(self.parent().po.data_list), self.parent().po.bot[i] - self.parent().po.top[i] + add_to_c, self.parent().po.right[i] - self.parent().po.left[i] + add_to_c),
+                    dtype=np.uint8)
+            else:
+                self.parent().po.converted_video = np.zeros(
+                    (len(self.parent().po.data_list), self.parent().po.bot[i] - self.parent().po.top[i] + add_to_c, self.parent().po.right[i] - self.parent().po.left[i] + add_to_c),
+                    dtype=float)
+            if not self.parent().po.vars['already_greyscale']:
+                self.parent().po.visu = np.zeros((len(self.parent().po.data_list), self.parent().po.bot[i] - self.parent().po.top[i] + add_to_c,
+                                   self.parent().po.right[i] - self.parent().po.left[i] + add_to_c, 3), dtype=np.uint8)
+                if self.parent().po.vars['convert_for_motion']['logical'] != 'None':
+                    if self.parent().po.vars['lose_accuracy_to_save_memory']:
+                        self.parent().po.converted_video2 = np.zeros((len(self.parent().po.data_list), self.parent().po.bot[i] - self.parent().po.top[i] + add_to_c,
+                                                       self.parent().po.right[i] - self.parent().po.left[i] + add_to_c), dtype=np.uint8)
+                    else:
+                        self.parent().po.converted_video2 = np.zeros((len(self.parent().po.data_list), self.parent().po.bot[i] - self.parent().po.top[i] + add_to_c,
+                                                       self.parent().po.right[i] - self.parent().po.left[i] + add_to_c), dtype=float)
+                first_dict = TDict()
+                second_dict = TDict()
+                c_spaces = []
+                for k, v in self.parent().po.vars['convert_for_motion'].items():
+                     if k != 'logical' and v.sum() > 0:
+                        if k[-1] != '2':
+                            first_dict[k] = v
+                            c_spaces.append(k)
+                        else:
+                            second_dict[k[:-1]] = v
+                            c_spaces.append(k[:-1])
+            prev_img = None
+            background = None
+            background2 = None
+            pat_tracker = PercentAndTimeTracker(self.parent().po.vars['img_number'])
+            for image_i, image_name in enumerate(self.parent().po.data_list):
+                current_percentage, eta = pat_tracker.get_progress()
+                is_landscape = self.parent().po.first_image.image.shape[0] < self.parent().po.first_image.image.shape[1]
+                img = read_and_rotate(image_name, prev_img, self.parent().po.all['raw_images'], is_landscape)
+                # img = self.parent().po.videos.read_and_rotate(image_name, prev_img)
+                prev_img = deepcopy(img)
+                if self.parent().po.first_image.cropped:
+                    img = img[self.parent().po.first_image.crop_coord[0]:self.parent().po.first_image.crop_coord[1],
+                          self.parent().po.first_image.crop_coord[2]:self.parent().po.first_image.crop_coord[3], :]
+                img = img[self.parent().po.top[arena - 1]: (self.parent().po.bot[arena - 1] + add_to_c),
+                      self.parent().po.left[arena - 1]: (self.parent().po.right[arena - 1] + add_to_c), :]
+
+                self.image_from_thread.emit({"message": f"Video loading: {current_percentage}%{eta}", "current_image": img})
+                if self.parent().po.vars['already_greyscale']:
+                    if self.parent().po.reduce_image_dim:
+                        self.parent().po.converted_video[image_i, ...] = img[:, :, 0]
+                    else:
+                        self.parent().po.converted_video[image_i, ...] = img
+                else:
+                    self.parent().po.visu[image_i, ...] = img
+
+                    if self.parent().po.vars['subtract_background']:
+                        background = self.parent().po.vars['background_list'][i]
+                        if self.parent().po.vars['convert_for_motion']['logical'] != 'None':
+                            background2 = self.parent().po.vars['background_list2'][i]
+                    greyscale_image, greyscale_image2 = generate_color_space_combination(img, c_spaces,
+                                                                                         first_dict,
+                                                                                         second_dict,background,background2,
+                                                                                         self.parent().po.vars[
+                                                                                             'lose_accuracy_to_save_memory'])
+                    self.parent().po.converted_video[image_i, ...] = greyscale_image
+                    if self.parent().po.vars['convert_for_motion']['logical'] != 'None':
+                        self.parent().po.converted_video2[image_i, ...] = greyscale_image2
+
+
+
+                    # csc = OneImageAnalysis(img)
+                    # if self.parent().po.vars['subtract_background']:
+                    #     if self.parent().po.vars['convert_for_motion']['logical'] != 'None':
+                    #         csc.generate_color_space_combination(c_spaces, first_dict, second_dict,
+                    #                                              self.parent().po.vars['background_list'][i],
+                    #                                              self.parent().po.vars['background_list2'][i])
+                    #     else:
+                    #         csc.generate_color_space_combination(c_spaces, first_dict, second_dict,
+                    #                                              self.parent().po.vars['background_list'][i], None)
+                    # else:
+                    #     csc.generate_color_space_combination(c_spaces, first_dict, second_dict, None, None)
+                    # # self.parent().po.converted_video[image_i, ...] = csc.image
+                    # if self.parent().po.vars['lose_accuracy_to_save_memory']:
+                    #     self.parent().po.converted_video[image_i, ...] = bracket_to_np.uint8_image_contrast(csc.image)
+                    # else:
+                    #     self.parent().po.converted_video[image_i, ...] = csc.image
+                    # if self.parent().po.vars['convert_for_motion']['logical'] != 'None':
+                    #     if self.parent().po.vars['lose_accuracy_to_save_memory']:
+                    #         self.parent().po.converted_video2[image_i, ...] = bracket_to_np.uint8_image_contrast(csc.image2)
+                    #     else:
+                    #         self.parent().po.converted_video2[image_i, ...] = csc.image2
+
+
+
+            # self.parent().po.load_one_arena(arena)
+            save_loaded_video = True
+            if self.parent().po.vars['already_greyscale']:
+                self.videos_in_ram = self.parent().po.converted_video
+            else:
+                if self.parent().po.vars['convert_for_motion']['logical'] == 'None':
+                    self.videos_in_ram = [self.parent().po.visu, deepcopy(self.parent().po.converted_video)]
+                else:
+                    self.videos_in_ram = [self.parent().po.visu, deepcopy(self.parent().po.converted_video), deepcopy(self.parent().po.converted_video2)]
+
+            # videos = [self.parent().po.video.copy(), self.parent().po.converted_video.copy()]
+        else:
+            logging.info(f"Starting to load arena n°{arena} from .npy saved file")
+            self.videos_in_ram = None
+        l = [i, arena, self.parent().po.vars, False, False, False, self.videos_in_ram]
+        self.parent().po.motion = MotionAnalysis(l)
+        r = weakref.ref(self.parent().po.motion)
+
+        if self.videos_in_ram is None:
+            self.parent().po.converted_video = deepcopy(self.parent().po.motion.converted_video)
+            if self.parent().po.vars['convert_for_motion']['logical'] != 'None':
+                self.parent().po.converted_video2 = deepcopy(self.parent().po.motion.converted_video2)
+        self.parent().po.motion.get_origin_shape()
+
+        if self.parent().po.motion.dims[0] >= 40:
+            step = self.parent().po.motion.dims[0] // 20
+        else:
+            step = 1
+        if self.parent().po.motion.start >= (self.parent().po.motion.dims[0] - step - 1):
+            self.parent().po.motion.start = None
+        else:
+            self.parent().po.motion.get_covering_duration(step)
+        self.when_loading_finished.emit(save_loaded_video)
+
+        if self.parent().po.motion.visu is None:
+            visu = self.parent().po.motion.converted_video
+            visu -= np.min(visu)
+            visu = 255 * (visu / np.max(visu))
+            visu = np.round(visu).astype(np.uint8)
+            if len(visu.shape) == 3:
+                visu = np.stack((visu, visu, visu), axis=3)
+            self.parent().po.motion.visu = visu
+
+    def detection(self):
+        self.message_from_thread_starting.emit(f"Quick video segmentation")
+        self.parent().po.motion.converted_video = deepcopy(self.parent().po.converted_video)
+        if self.parent().po.vars['convert_for_motion']['logical'] != 'None':
+            self.parent().po.motion.converted_video2 = deepcopy(self.parent().po.converted_video2)
+        # self.parent().po.motion.detection(compute_all_possibilities=True)
+        self.parent().po.motion.detection(compute_all_possibilities=self.parent().po.all['compute_all_options'])
+        if self.parent().po.all['compute_all_options']:
+            self.parent().po.computed_video_options = np.ones(5, bool)
+        else:
+            self.parent().po.computed_video_options = np.zeros(5, bool)
+            self.parent().po.computed_video_options[self.parent().po.all['video_option']] = True
+        # if self.parent().po.vars['color_number'] > 2:
+
+    def post_processing(self):
+        self.parent().po.motion.smoothed_video = None
+        # if self.parent().po.vars['already_greyscale']:
+        #     if self.parent().po.vars['convert_for_motion']['logical'] == 'None':
+        #         self.videos_in_ram = self.parent().po.converted_video
+        #     else:
+        #         self.videos_in_ram = self.parent().po.converted_video, self.parent().po.converted_video2
+        # else:
+        #     if self.parent().po.vars['convert_for_motion']['logical'] == 'None':
+        #         videos_in_ram = self.parent().po.visu, self.parent().po.converted_video
+        #     else:
+        #         videos_in_ram = self.parent().po.visu, self.parent().po.converted_video, \
+        #                         self.parent().po.converted_video2
+
+        if self.parent().po.vars['color_number'] > 2:
+            analyses_to_compute = [0]
+        else:
+            if self.parent().po.all['compute_all_options']:
+                analyses_to_compute = np.arange(5)
+            else:
+                logging.info(f"option: {self.parent().po.all['video_option']}")
+                analyses_to_compute = [self.parent().po.all['video_option']]
+        time_parameters = [self.parent().po.motion.start, self.parent().po.motion.step,
+                           self.parent().po.motion.lost_frames, self.parent().po.motion.substantial_growth]
+
+        args = [self.parent().po.all['arena'] - 1, self.parent().po.all['arena'], self.parent().po.vars,
+                False, False, False, self.videos_in_ram]
+        if self.parent().po.vars['do_fading']:
+            self.parent().po.newly_explored_area = np.zeros((self.parent().po.motion.dims[0], 5), np.unp.int64)
+        for seg_i in analyses_to_compute:
+            analysis_i = MotionAnalysis(args)
+            r = weakref.ref(analysis_i)
+            analysis_i.segmentation = np.zeros(analysis_i.converted_video.shape[:3], dtype=np.uint8)
+            if self.parent().po.all['compute_all_options']:
+                if seg_i == 0:
+                    analysis_i.segmentation = self.parent().po.motion.segmentation
+                else:
+                    if seg_i == 1:
+                        mask = self.parent().po.motion.luminosity_segmentation
+                    elif seg_i == 2:
+                        mask = self.parent().po.motion.gradient_segmentation
+                    elif seg_i == 3:
+                        mask = self.parent().po.motion.logical_and
+                    elif seg_i == 4:
+                        mask = self.parent().po.motion.logical_or
+                    analysis_i.segmentation[mask[0], mask[1], mask[2]] = 1
+            else:
+                if self.parent().po.computed_video_options[self.parent().po.all['video_option']]:
+                    analysis_i.segmentation = self.parent().po.motion.segmentation
+
+            analysis_i.start = time_parameters[0]
+            analysis_i.step = time_parameters[1]
+            analysis_i.lost_frames = time_parameters[2]
+            analysis_i.substantial_growth = time_parameters[3]
+            analysis_i.origin_idx = self.parent().po.motion.origin_idx
+            analysis_i.initialize_post_processing()
+            analysis_i.t = analysis_i.start
+            # print_progress = ForLoopCounter(self.start)
+
+            while self._isRunning and analysis_i.t < analysis_i.binary.shape[0]:
+                # analysis_i.update_shape(True)
+                analysis_i.update_shape(False)
+                contours = np.nonzero(
+                    cv2.morphologyEx(analysis_i.binary[analysis_i.t - 1, :, :], cv2.MORPH_GRADIENT, cross_33))
+                current_image = deepcopy(self.parent().po.motion.visu[analysis_i.t - 1, :, :, :])
+                current_image[contours[0], contours[1], :] = self.parent().po.vars['contour_color']
+                self.image_from_thread.emit(
+                    {"message": f"Tracking option n°{seg_i + 1}. Image number: {analysis_i.t - 1}",
+                     "current_image": current_image})
+            if analysis_i.start is None:
+                analysis_i.binary = np.repeat(np.expand_dims(analysis_i.origin, 0),
+                                           analysis_i.converted_video.shape[0], axis=0)
+                if self.parent().po.vars['color_number'] > 2:
+                    self.message_from_thread_starting.emit(
+                        f"Failed to detect motion. Redo image analysis (with only 2 colors?)")
+                else:
+                    self.message_from_thread_starting.emit(f"Tracking option n°{seg_i + 1} failed to detect motion")
+
+            if self.parent().po.all['compute_all_options']:
+                if seg_i == 0:
+                    self.parent().po.motion.segmentation = analysis_i.binary
+                elif seg_i == 1:
+                    self.parent().po.motion.luminosity_segmentation = np.nonzero(analysis_i.binary)
+                elif seg_i == 2:
+                    self.parent().po.motion.gradient_segmentation = np.nonzero(analysis_i.binary)
+                elif seg_i == 3:
+                    self.parent().po.motion.logical_and = np.nonzero(analysis_i.binary)
+                elif seg_i == 4:
+                    self.parent().po.motion.logical_or = np.nonzero(analysis_i.binary)
+            else:
+                self.parent().po.motion.segmentation = analysis_i.binary
+
+        # self.message_from_thread_starting.emit("If there are problems, change some parameters and try again")
+        self.when_detection_finished.emit("Post processing done, read to see the result")
+
+
+
+class VideoReaderThread(QtCore.QThread):
+    message_from_thread = QtCore.Signal(dict)
+
+    def __init__(self, parent=None):
+        super(VideoReaderThread, self).__init__(parent)
+        self.setParent(parent)
+
+    def run(self):
+        video_analysis = deepcopy(self.parent().po.motion.visu)
+        self.message_from_thread.emit(
+            {"current_image": video_analysis[0, ...], "message": f"Video preparation, wait..."})
+        if self.parent().po.load_quick_full > 0:
+
+            if self.parent().po.all['compute_all_options']:
+                if self.parent().po.all['video_option'] == 0:
+                    video_mask = self.parent().po.motion.segmentation
+                else:
+                    if self.parent().po.all['video_option'] == 1:
+                        mask = self.parent().po.motion.luminosity_segmentation
+                    elif self.parent().po.all['video_option'] == 2:
+                        mask = self.parent().po.motion.gradient_segmentation
+                    elif self.parent().po.all['video_option'] == 3:
+                        mask = self.parent().po.motion.logical_and
+                    elif self.parent().po.all['video_option'] == 4:
+                        mask = self.parent().po.motion.logical_or
+                    video_mask = np.zeros(self.parent().po.motion.dims[:3], dtype=np.uint8)
+                    video_mask[mask[0], mask[1], mask[2]] = 1
+            else:
+                video_mask = np.zeros(self.parent().po.motion.dims[:3], dtype=np.uint8)
+                if self.parent().po.computed_video_options[self.parent().po.all['video_option']]:
+                    video_mask = self.parent().po.motion.segmentation
+
+            if self.parent().po.load_quick_full == 1:
+                video_mask = np.cumsum(video_mask.astype(np.uint32), axis=0)
+                video_mask[video_mask > 0] = 1
+                video_mask = video_mask.astype(np.uint8)
+        logging.info(f"sum: {video_mask.sum()}")
+        # timings = genfromtxt("timings.csv")
+        for t in np.arange(self.parent().po.motion.dims[0]):
+            mask = cv2.morphologyEx(video_mask[t, ...], cv2.MORPH_GRADIENT, cross_33)
+            mask = np.stack((mask, mask, mask), axis=2)
+            # current_image[current_image > 0] = self.parent().po.vars['contour_color']
+            current_image = deepcopy(video_analysis[t, ...])
+            current_image[mask > 0] = self.parent().po.vars['contour_color']
+            self.message_from_thread.emit(
+                {"current_image": current_image, "message": f"Reading in progress... Image number: {t}"}) #, "time": timings[t]
+            time.sleep(1 / 50)
+        self.message_from_thread.emit({"current_image": current_image, "message": ""})#, "time": timings[t]
+
+
+class ChangeOneRepResultThread(QtCore.QThread):
+    message_from_thread = QtCore.Signal(str)
+
+    def __init__(self, parent=None):
+        super(ChangeOneRepResultThread, self).__init__(parent)
+        self.setParent(parent)
+
+    def run(self):
+        self.message_from_thread.emit(
+            f"Arena n°{self.parent().po.all['arena']}: modifying its results...")
+        # self.parent().po.motion2 = deepcopy(self.parent().po.motion)
+        if self.parent().po.motion.start is None:
+            self.parent().po.motion.binary = np.repeat(np.expand_dims(self.parent().po.motion.origin, 0),
+                                                     self.parent().po.motion.converted_video.shape[0], axis=0).astype(np.uint8)
+        else:
+            if self.parent().po.all['compute_all_options']:
+                if self.parent().po.all['video_option'] == 0:
+                    self.parent().po.motion.binary = self.parent().po.motion.segmentation
+                else:
+                    if self.parent().po.all['video_option'] == 1:
+                        mask = self.parent().po.motion.luminosity_segmentation
+                    elif self.parent().po.all['video_option'] == 2:
+                        mask = self.parent().po.motion.gradient_segmentation
+                    elif self.parent().po.all['video_option'] == 3:
+                        mask = self.parent().po.motion.logical_and
+                    elif self.parent().po.all['video_option'] == 4:
+                        mask = self.parent().po.motion.logical_or
+                    self.parent().po.motion.binary = np.zeros(self.parent().po.motion.dims, dtype=np.uint8)
+                    self.parent().po.motion.binary[mask[0], mask[1], mask[2]] = 1
+            else:
+                self.parent().po.motion.binary = np.zeros(self.parent().po.motion.dims[:3], dtype=np.uint8)
+                if self.parent().po.computed_video_options[self.parent().po.all['video_option']]:
+                    self.parent().po.motion.binary = self.parent().po.motion.segmentation
+
+        if self.parent().po.vars['do_fading']:
+            self.parent().po.motion.newly_explored_area = self.parent().po.newly_explored_area[:, self.parent().po.all['video_option']]
+        self.parent().po.motion.max_distance = 9 * self.parent().po.vars['detection_range_factor']
+        self.parent().po.motion.get_descriptors_from_binary(release_memory=False)
+        self.parent().po.motion.detect_growth_transitions()
+        self.parent().po.motion.networks_detection(False)
+        self.parent().po.motion.study_cytoscillations(False)
+        self.parent().po.motion.fractal_descriptions()
+        self.parent().po.motion.get_descriptors_summary()
+        self.parent().po.motion.change_results_of_one_arena()
+        self.parent().po.motion = None
+        # self.parent().po.motion = None
+        self.message_from_thread.emit("")
+
+
+class WriteVideoThread(QtCore.QThread):
+    # message_from_thread_in_thread = QtCore.Signal(bool)
+    def __init__(self, parent=None):
+        super(WriteVideoThread, self).__init__(parent)
+        self.setParent(parent)
+
+    def run(self):
+        # self.message_from_thread_in_thread.emit({True})
+        arena = self.parent().po.all['arena']
+        if not self.parent().po.vars['already_greyscale']:
+            write_video(self.parent().po.visu, f'ind_{arena}.npy')
+        else:
+            write_video(self.parent().po.converted_video, f'ind_{arena}.npy')
+
+
+class RunAllThread(QtCore.QThread):
+    message_from_thread = QtCore.Signal(str)
+    image_from_thread = QtCore.Signal(dict)
+
+    def __init__(self, parent=None):
+        super(RunAllThread, self).__init__(parent)
+        self.setParent(parent)
+
+    def run(self):
+        analysis_status = {"continue": True, "message": ""}
+        message = self.set_current_folder(0)
+
+        if self.parent().po.first_exp_ready_to_run:
+
+            self.message_from_thread.emit(message + ": Write videos...")
+            if not self.parent().po.vars['several_blob_per_arena'] and self.parent().po.sample_number != len(self.parent().po.bot):
+                analysis_status["continue"] = False
+                analysis_status["message"] = f"Wrong specimen number: redo the first image analysis."
+                self.message_from_thread.emit(f"Wrong specimen number: restart Cellects and do another analysis.")
+            else:
+                analysis_status = self.run_video_writing(message)
+                if analysis_status["continue"]:
+                    self.message_from_thread.emit(message + ": Analyse all videos...")
+                    analysis_status = self.run_motion_analysis(message)
+                if analysis_status["continue"]:
+                    if self.parent().po.all['folder_number'] > 1:
+                        self.parent().po.all['folder_list'] = self.parent().po.all['folder_list'][1:]
+                        self.parent().po.all['sample_number_per_folder'] = self.parent().po.all['sample_number_per_folder'][1:]
+        else:
+            self.parent().po.look_for_data()
+
+        if analysis_status["continue"] and (not self.parent().po.first_exp_ready_to_run or self.parent().po.all['folder_number'] > 1):
+            folder_number = np.max((len(self.parent().po.all['folder_list']), 1))
+
+            for exp_i in np.arange(folder_number):
+                if len(self.parent().po.all['folder_list']) > 0:
+                    logging.info(self.parent().po.all['folder_list'][exp_i])
+                self.parent().po.first_im = None
+                self.parent().po.first_image = None
+                self.parent().po.last_im = None
+                self.parent().po.last_image = None
+                self.parent().po.videos = None
+                self.parent().po.top = None
+
+                message = self.set_current_folder(exp_i)
+                self.message_from_thread.emit(f'{message}, pre-processing...')
+                self.parent().po.load_data_to_run_cellects_quickly()
+                if not self.parent().po.first_exp_ready_to_run:
+                    analysis_status = self.pre_processing()
+                if analysis_status["continue"]:
+                    self.message_from_thread.emit(message + ": Write videos from images before analysis...")
+                    if not self.parent().po.vars['several_blob_per_arena'] and self.parent().po.sample_number != len(self.parent().po.bot):
+                        self.message_from_thread.emit(f"Wrong specimen number: first image analysis is mandatory.")
+                        analysis_status["continue"] = False
+                        analysis_status["message"] = f"Wrong specimen number: first image analysis is mandatory."
+                    else:
+                        analysis_status = self.run_video_writing(message)
+                        if analysis_status["continue"]:
+                            self.message_from_thread.emit(message + ": Starting analysis...")
+                            analysis_status = self.run_motion_analysis(message)
+
+                if not analysis_status["continue"]:
+                    # self.message_from_thread.emit(analysis_status["message"])
+                    break
+                # if not continue_analysis:
+                #     self.message_from_thread.emit(f"Error: wrong folder or parameters")
+                #     break
+                # if not enough_memory:
+                #     self.message_from_thread.emit(f"Error: not enough memory")
+                #     break
+                print(self.parent().po.vars['convert_for_motion'])
+        if analysis_status["continue"]:
+            if self.parent().po.all['folder_number'] > 1:
+                self.message_from_thread.emit(f"Exp {self.parent().po.all['folder_list'][0]} to {self.parent().po.all['folder_list'][-1]} analyzed.")
+            else:
+                curr_path = reduce_path_len(self.parent().po.all['global_pathway'], 6, 10)
+                self.message_from_thread.emit(f'Exp {curr_path}, analyzed.')
+        else:
+            logging.error(message + " " + analysis_status["message"])
+            self.message_from_thread.emit(message + " " + analysis_status["message"])
+
+    def set_current_folder(self, exp_i):
+        if self.parent().po.all['folder_number'] > 1:
+            logging.info(f"Use {self.parent().po.all['folder_list'][exp_i]} folder")
+
+            message = f"{str(self.parent().po.all['global_pathway'])[:6]} ... {self.parent().po.all['folder_list'][exp_i]}"
+            self.parent().po.update_folder_id(self.parent().po.all['sample_number_per_folder'][exp_i],
+                                              self.parent().po.all['folder_list'][exp_i])
+        else:
+            message = reduce_path_len(self.parent().po.all['global_pathway'], 6, 10)
+            logging.info(f"Use {message} folder")
+            self.parent().po.update_folder_id(self.parent().po.all['first_folder_sample_number'])
+        return message
+
+    def pre_processing(self):
+        analysis_status = {"continue": True, "message": ""}
+        logging.info("Pre-processing has started")
+        if len(self.parent().po.data_list) > 0:
+            self.parent().po.get_first_image()
+            self.parent().po.fast_image_segmentation(True)
+            self.parent().po.cropping(is_first_image=True)
+            self.parent().po.get_average_pixel_size()
+            try:
+                analysis_status = self.parent().po.delineate_each_arena()
+            except ValueError:
+                analysis_status[
+                    "message"] = f"Failed to detect the right cell(s) number: the first image analysis is mandatory."
+                analysis_status["continue"] = False
+
+            if analysis_status["continue"]:
+                self.parent().po.data_to_save['exif'] = True
+                self.parent().po.save_data_to_run_cellects_quickly()
+                self.parent().po.data_to_save['exif'] = False
+                # self.parent().po.extract_exif()
+                self.parent().po.get_background_to_subtract()
+                if len(self.parent().po.vars['analyzed_individuals']) != len(self.parent().po.top):
+                    analysis_status["message"] = f"Failed to detect the right cell(s) number: the first image analysis is mandatory."
+                    analysis_status["continue"] = False
+                elif self.parent().po.top is None and self.parent().imageanalysiswindow.manual_delineation_flag:
+                    analysis_status["message"] = f"Auto video delineation failed, use manual delineation tool"
+                    analysis_status["continue"] = False
+                else:
+                    self.parent().po.get_origins_and_backgrounds_lists()
+                    self.parent().po.get_last_image()
+                    self.parent().po.fast_image_segmentation(is_first_image=False)
+                    self.parent().po.find_if_lighter_backgnp.round()
+            return analysis_status
+        else:
+            analysis_status["message"] = f"Wrong folder or parameters"
+            analysis_status["continue"] = False
+            return analysis_status
+
+    def run_video_writing(self, message):
+        analysis_status = {"continue": True, "message": ""}
+        look_for_existing_videos = glob('ind_' + '*' + '.npy')
+        there_already_are_videos = len(look_for_existing_videos) == len(self.parent().po.vars['analyzed_individuals'])
+        logging.info(f"{len(look_for_existing_videos)} .npy video files found for {len(self.parent().po.vars['analyzed_individuals'])} arenas to analyze")
+        do_write_videos = not there_already_are_videos or (
+                there_already_are_videos and self.parent().po.all['overwrite_unaltered_videos'])
+        if do_write_videos:
+            logging.info(f"Starting video writing")
+            # self.videos.write_videos_as_np_arrays(self.data_list, self.vars['convert_for_motion'], in_colors=self.vars['save_in_colors'])
+            in_colors = not self.parent().po.vars['already_greyscale']
+            self.parent().po.videos = OneVideoPerBlob(self.parent().po.first_image,
+                                                      self.parent().po.starting_blob_hsize_in_pixels,
+                                                      self.parent().po.all['raw_images'])
+            self.parent().po.videos.left = self.parent().po.left
+            self.parent().po.videos.right = self.parent().po.right
+            self.parent().po.videos.top = self.parent().po.top
+            self.parent().po.videos.bot = self.parent().po.bot
+            self.parent().po.videos.first_image.shape_number = self.parent().po.sample_number
+            bunch_nb, video_nb_per_bunch, sizes, video_bunch, vid_names, rom_memory_required, analysis_status, remaining = self.parent().po.videos.prepare_video_writing(
+                self.parent().po.data_list, self.parent().po.vars['min_ram_free'], in_colors)
+            if analysis_status["continue"]:
+                # Check that there is enough available RAM for one video par bunch and ROM for all videos
+                if video_nb_per_bunch > 0 and rom_memory_required is None:
+                    pat_tracker1 = PercentAndTimeTracker(bunch_nb * self.parent().po.vars['img_number'])
+                    pat_tracker2 = PercentAndTimeTracker(len(self.parent().po.vars['analyzed_individuals']))
+                    arena_percentage = 0
+                    for bunch in np.arange(bunch_nb):
+                        # Update the labels of arenas and the video_bunch to write
+                        if bunch == (bunch_nb - 1) and remaining > 0:
+                            arena = np.arange(bunch * video_nb_per_bunch, bunch * video_nb_per_bunch + remaining)
+                        else:
+                            arena = np.arange(bunch * video_nb_per_bunch, (bunch + 1) * video_nb_per_bunch)
+                        if self.parent().po.videos.use_list_of_vid:
+                            video_bunch = [np.zeros(sizes[i, :], dtype=np.uint8) for i in arena]
+                        else:
+                            video_bunch = np.zeros(np.append(sizes[0, :], len(arena)), dtype=np.uint8)
+                        prev_img = None
+                        images_done = bunch * self.parent().po.vars['img_number']
+                        for image_i, image_name in enumerate(self.parent().po.data_list):
+                            image_percentage, remaining_time = pat_tracker1.get_progress(image_i + images_done)
+                            self.message_from_thread.emit(message + f" Step 1/2: Video writing ({np.round((image_percentage + arena_percentage) / 2, 2)}%)")
+                            if not os.path.exists(image_name):
+                                raise FileNotFoundError(image_name)
+                            img = self.parent().po.videos.read_and_rotate(image_name, prev_img)
+                            prev_img = deepcopy(img)
+                            if self.parent().po.vars['already_greyscale'] and self.parent().po.reduce_image_dim:
+                                img = img[:, :, 0]
+
+                            for arena_i, arena_name in enumerate(arena):
+                                try:
+                                    sub_img = img[self.parent().po.top[arena_name]: (self.parent().po.bot[arena_name] + 1),
+                                              self.parent().po.left[arena_name]: (self.parent().po.right[arena_name] + 1), ...]
+                                    if self.parent().po.videos.use_list_of_vid:
+                                        video_bunch[arena_i][image_i, ...] = sub_img
+                                    else:
+                                        if len(video_bunch.shape) == 5:
+                                            video_bunch[image_i, :, :, :, arena_i] = sub_img
+                                        else:
+                                            video_bunch[image_i, :, :, arena_i] = sub_img
+                                except ValueError:
+                                    analysis_status["message"] = f"One (or more) image has a different size (restart)"
+                                    analysis_status["continue"] = False
+                                    logging.info(f"In the {message} folder: one (or more) image has a different size (restart)")
+                                    break
+                            if not analysis_status["continue"]:
+                                break
+                        if not analysis_status["continue"]:
+                            break
+                        if analysis_status["continue"]:
+                            for arena_i, arena_name in enumerate(arena):
+                                try:
+                                    arena_percentage, eta = pat_tracker2.get_progress()
+                                    self.message_from_thread.emit(message + f" Step 1/2: Video writing ({np.round((image_percentage + arena_percentage) / 2, 2)}%)")# , ETA {remaining_time}
+                                    if self.parent().po.videos.use_list_of_vid:
+                                        np.save(vid_names[arena_name], video_bunch[arena_i])
+                                    else:
+                                        if len(video_bunch.shape) == 5:
+                                            np.save(vid_names[arena_name], video_bunch[:, :, :, :, arena_i])
+                                        else:
+                                            np.save(vid_names[arena_name], video_bunch[:, :, :, arena_i])
+                                except OSError:
+                                    self.message_from_thread.emit(message + f"full disk memory, clear space and retry")
+                        logging.info(f"Bunch n°{bunch + 1} over {bunch_nb} saved.")
+                    logging.info("When they exist, do not overwrite unaltered video")
+                    self.parent().po.all['overwrite_unaltered_videos'] = False
+                    self.parent().po.save_variable_dict()
+                    self.parent().po.save_data_to_run_cellects_quickly()
+                    analysis_status["message"] = f"Video writing complete."
+                    if self.parent().po.videos is not None:
+                        del self.parent().po.videos
+                    return analysis_status
+                else:
+                    analysis_status["continue"] = False
+                    if video_nb_per_bunch == 0:
+                        memory_diff = self.parent().po.update_available_core_nb()
+                        ram_message = f"{memory_diff}GB of additional RAM"
+                    if rom_memory_required is not None:
+                        rom_message = f"at least {rom_memory_required}GB of free ROM"
+
+                    if video_nb_per_bunch == 0 and rom_memory_required is not None:
+                        analysis_status["message"] = f"Requires {ram_message} and {rom_message} to run"
+                        # self.message_from_thread.emit(f"Analyzing {message} requires {ram_message} and {rom_message} to run")
+                    elif video_nb_per_bunch == 0:
+                        analysis_status["message"] = f"Requires {ram_message} to run"
+                        # self.message_from_thread.emit(f"Analyzing {message} requires {ram_message} to run")
+                    elif rom_memory_required is not None:
+                        analysis_status["message"] = f"Requires {rom_message} to run"
+                        # self.message_from_thread.emit(f"Analyzing {message} requires {rom_message} to run")
+                    logging.info(f"Cellects is not writing videos: insufficient memory")
+                    return analysis_status
+            else:
+                return analysis_status
+
+
+        else:
+            logging.info(f"Cellects is not writing videos: unnecessary")
+            analysis_status["message"] = f"Cellects is not writing videos: unnecessary"
+            return analysis_status
+
+    def run_motion_analysis(self, message):
+        analysis_status = {"continue": True, "message": ""}
+        logging.info(f"Starting motion analysis with the detection method n°{self.parent().po.all['video_option']}")
+        self.parent().po.instantiate_tables()
+        try:
+            memory_diff = self.parent().po.update_available_core_nb()
+            if self.parent().po.cores > 0: # i.e. enough memory
+                if not self.parent().po.all['do_multiprocessing'] or self.parent().po.cores == 1:
+                    self.message_from_thread.emit(f"{message} Step 2/2: Video analysis")
+                    logging.info("fStarting sequential analysis")
+                    tiii = default_timer()
+                    pat_tracker = PercentAndTimeTracker(len(self.parent().po.vars['analyzed_individuals']))
+                    for i, arena in enumerate(self.parent().po.vars['analyzed_individuals']):
+
+                        l = [i, arena, self.parent().po.vars, True, True, False, None]
+                        # l = [0, 1, self.parent().po.vars, True, False, False, None]
+                        analysis_i = MotionAnalysis(l)
+                        r = weakref.ref(analysis_i)
+                        if not self.parent().po.vars['several_blob_per_arena']:
+                            # Save basic statistics
+                            self.parent().po.update_one_row_per_arena(i, analysis_i.one_descriptor_per_arena)
+
+
+                            # Save descriptors in long_format
+                            self.parent().po.update_one_row_per_frame(i * self.parent().po.vars['img_number'], arena * self.parent().po.vars['img_number'], analysis_i.one_row_per_frame)
+                            
+                            # Save cytosol_oscillations
+                        if not pd.isna(analysis_i.one_descriptor_per_arena["first_move"]):
+                            if self.parent().po.vars['oscilacyto_analysis']:
+                                oscil_i = pd.DataFrame(
+                                    np.c_[np.repeat(arena,
+                                              analysis_i.clusters_final_data.shape[0]), analysis_i.clusters_final_data],
+                                    columns=['arena', 'mean_pixel_period', 'phase', 'cluster_size', 'edge_distance', 'coord_y', 'coord_x'])
+                                if self.parent().po.one_row_per_oscillating_cluster is None:
+                                    self.parent().po.one_row_per_oscillating_cluster = oscil_i
+                                else:
+                                    self.parent().po.one_row_per_oscillating_cluster = pd.concat((self.parent().po.one_row_per_oscillating_cluster, oscil_i))
+                                
+                        # Save efficiency visualization
+                        self.parent().po.add_analysis_visualization_to_first_and_last_images(i, analysis_i.efficiency_test_1,
+                                                                                 analysis_i.efficiency_test_2)
+                        # Emit message to the interface
+                        current_percentage, eta = pat_tracker.get_progress()
+                        self.image_from_thread.emit({"current_image": self.parent().po.last_image.bgr,
+                                                     "message": f"{message} Step 2/2: analyzed {arena} out of {len(self.parent().po.vars['analyzed_individuals'])} arenas ({current_percentage}%){eta}"})
+                        del analysis_i
+                    logging.info(f"Sequential analysis lasted {(default_timer() - tiii)/ 60} minutes")
+                else:
+                    self.message_from_thread.emit(
+                        f"{message}, Step 2/2:  Analyse all videos using {self.parent().po.cores} cores...")
+
+                    logging.info("fStarting analysis in parallel")
+
+                    # new
+                    tiii = default_timer()
+                    arena_number = len(self.parent().po.vars['analyzed_individuals'])
+                    self.advance = 0
+                    self.pat_tracker = PercentAndTimeTracker(len(self.parent().po.vars['analyzed_individuals']),
+                                                        core_number=self.parent().po.cores)
+
+                    fair_core_workload = arena_number // self.parent().po.cores
+                    cores_with_1_more = arena_number % self.parent().po.cores
+                    EXTENTS_OF_SUBRANGES = []
+                    bound = 0
+                    parallel_organization = [fair_core_workload + 1 for _ in range(cores_with_1_more)] + [fair_core_workload for _ in range(self.parent().po.cores - cores_with_1_more)]
+                    # Emit message to the interface
+                    self.image_from_thread.emit({"current_image": self.parent().po.last_image.bgr,
+                                                 "message": f"{message} Step 2/2: Analysis running on {self.parent().po.cores} CPU cores"})
+                    for i, extent_size in enumerate(parallel_organization):
+                        EXTENTS_OF_SUBRANGES.append((bound, bound := bound + extent_size))
+
+                    try:
+                        PROCESSES = []
+                        subtotals = Manager().Queue()# Queue()
+                        for extent in EXTENTS_OF_SUBRANGES:
+                            # print(extent)
+                            p = Process(target=motion_analysis_process, args=(extent[0], extent[1], self.parent().po.vars, subtotals))
+                            p.start()
+                            PROCESSES.append(p)
+
+                        for p in PROCESSES:
+                            p.join()
+
+                        self.message_from_thread.emit(f"{message}, Step 2/2:  Saving all results...")
+                        for i in range(subtotals.qsize()):
+                            grouped_results = subtotals.get()
+                            for j, results_i in enumerate(grouped_results):
+                                if not self.parent().po.vars['several_blob_per_arena']:
+                                    # Save basic statistics
+                                    self.parent().po.update_one_row_per_arena(results_i['i'], results_i['one_row_per_arena'])
+                                    # Save descriptors in long_format
+                                    self.parent().po.update_one_row_per_frame(results_i['i'] * self.parent().po.vars['img_number'],
+                                                                              results_i['arena'] * self.parent().po.vars['img_number'],
+                                                                              results_i['one_row_per_frame'])
+                                if not pd.isna(results_i['first_move']):
+                                    # Save cytosol_oscillations
+                                    if self.parent().po.vars['oscilacyto_analysis']:
+                                        if self.parent().po.one_row_per_oscillating_cluster is None:
+                                            self.parent().po.one_row_per_oscillating_cluster = results_i['one_row_per_oscillating_cluster']
+                                        else:
+                                            self.parent().po.one_row_per_oscillating_cluster = pd.concat((self.parent().po.one_row_per_oscillating_cluster, results_i['one_row_per_oscillating_cluster']))
+                                        
+                                # Save efficiency visualization
+                                self.parent().po.add_analysis_visualization_to_first_and_last_images(results_i['i'], results_i['efficiency_test_1'],
+                                                                                         results_i['efficiency_test_2'])
+                        self.image_from_thread.emit(
+                            {"current_image": self.parent().po.last_image.bgr,
+                             "message": f"{message} Step 2/2: analyzed {len(self.parent().po.vars['analyzed_individuals'])} out of {len(self.parent().po.vars['analyzed_individuals'])} arenas ({100}%)"})
+
+                        logging.info(f"Parallel analysis lasted {(default_timer() - tiii)/ 60} minutes")
+                    except MemoryError:
+                        analysis_status["continue"] = False
+                        analysis_status["message"] = f"Not enough memory, reduce the core number for parallel analysis"
+                        self.message_from_thread.emit(f"Analyzing {message} requires to reduce the core number for parallel analysis")
+                        return analysis_status
+                self.parent().po.save_tables()
+                return analysis_status
+            else:
+                analysis_status["continue"] = False
+                analysis_status["message"] = f"Requires an additional {memory_diff}GB of RAM to run"
+                self.message_from_thread.emit(f"Analyzing {message} requires an additional {memory_diff}GB of RAM to run")
+                return analysis_status
+        except MemoryError:
+            analysis_status["continue"] = False
+            analysis_status["message"] = f"Requires additional memory to run"
+            self.message_from_thread.emit(f"Analyzing {message} requires additional memory to run")
+            return analysis_status
+
+
+def motion_analysis_process(lower_bound: int, upper_bound: int, vars: dict, subtotals: Queue) -> None:
+    grouped_results = []
+    for i in range(lower_bound, upper_bound):
+        analysis_i = MotionAnalysis([i, i + 1, vars, True, True, False, None])
+        r = weakref.ref(analysis_i)
+        results_i = dict()
+        results_i['arena'] = analysis_i.one_descriptor_per_arena['arena']
+        results_i['i'] = analysis_i.one_descriptor_per_arena['arena'] - 1
+        arena = results_i['arena']
+        i = arena - 1
+        if not vars['several_blob_per_arena']:
+            # Save basic statistics
+            results_i['one_row_per_arena'] = analysis_i.one_descriptor_per_arena
+            # Save descriptors in long_format
+            results_i['one_row_per_frame'] = analysis_i.one_row_per_frame
+            # Save cytosol_oscillations
+
+        results_i['first_move'] = analysis_i.one_descriptor_per_arena["first_move"]
+        if not pd.isna(analysis_i.one_descriptor_per_arena["first_move"]):
+            if vars['oscilacyto_analysis']:
+                results_i['clusters_final_data'] = analysis_i.clusters_final_data
+                results_i['one_row_per_oscillating_cluster'] = pd.DataFrame(
+                    np.c_[np.repeat(arena, analysis_i.clusters_final_data.shape[0]), analysis_i.clusters_final_data],
+                    columns=['arena', 'mean_pixel_period', 'phase', 'cluster_size', 'edge_distance', 'coord_y', 'coord_x'])
+            if vars['fractal_analysis']:
+                results_i['fractal_box_sizes'] = pd.DataFrame(analysis_i.fractal_boxes,
+                               columns=['arena', 'time', 'fractal_box_lengths', 'fractal_box_widths'])
+
+        # Save efficiency visualization
+        results_i['efficiency_test_1'] = analysis_i.efficiency_test_1
+        results_i['efficiency_test_2'] = analysis_i.efficiency_test_2
+        grouped_results.append(results_i)
+
+    subtotals.put(grouped_results)